From 3511dd75ac6911c03255f6c6c099811781002698 Mon Sep 17 00:00:00 2001
From: polyfloyd <floyd@polyfloyd.net>
Date: Fri, 5 Jan 2024 16:18:09 +0100
Subject: [PATCH] Initial commit

---
 .envrc                   |  1 +
 .gitignore               |  2 ++
 inflatinator/__main__.py | 16 +++++++++++++
 inflatinator/revbank.py  | 50 ++++++++++++++++++++++++++++++++++++++++
 inflatinator/scrapers.py | 37 +++++++++++++++++++++++++++++
 5 files changed, 106 insertions(+)
 create mode 100644 .envrc
 create mode 100644 .gitignore
 create mode 100644 inflatinator/__main__.py
 create mode 100644 inflatinator/revbank.py
 create mode 100644 inflatinator/scrapers.py

diff --git a/.envrc b/.envrc
new file mode 100644
index 0000000..175de89
--- /dev/null
+++ b/.envrc
@@ -0,0 +1 @@
+layout python
diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..533bcab
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,2 @@
+__pycache__
+/.direnv
diff --git a/inflatinator/__main__.py b/inflatinator/__main__.py
new file mode 100644
index 0000000..f8f7083
--- /dev/null
+++ b/inflatinator/__main__.py
@@ -0,0 +1,16 @@
+import revbank
+import sys
+
+
+def main(product_file):
+    with open(product_file, 'r') as fd:
+        src = fd.read()
+
+    new_src = revbank.update_product_pricings(src)
+
+    with open(product_file, 'w') as fd:
+        fd.write(new_src)
+
+
+if __name__ == '__main__':
+    main(sys.argv[1])
diff --git a/inflatinator/revbank.py b/inflatinator/revbank.py
new file mode 100644
index 0000000..8e3e38d
--- /dev/null
+++ b/inflatinator/revbank.py
@@ -0,0 +1,50 @@
+import re
+import scrapers
+from decimal import Decimal, ROUND_UP
+
+
+our_margin = Decimal('1.3')
+
+
+def find_product_details(vendor_and_sku):
+    [vendor, sku] = vendor_and_sku.split(':', 2)
+
+    if vendor == 'ah':
+        return scrapers.ah_get_by_sku(sku)
+
+    raise Exception(f'unknown vendor: {vendor}')
+
+
+def update_product_pricings(src):
+    lines = src.split('\n')
+
+    find_updatable = re.compile(r'#\s*(?P<sku>\S+)\s+(?P<units>\d+)x$')
+    find_aliases = re.compile(r'^(?P<aliases>\S+)')
+
+    lines_out = []
+
+    for line in lines:
+        m = find_updatable.search(line)
+        if not m:
+            lines_out.append(line)
+            continue
+
+        d = find_product_details(m['sku'])
+
+        product_aliases = set()
+        if not line.startswith('#'):
+            product_aliases = set(find_aliases.search(line)['aliases'].split(','))
+        product_aliases.add(d['ean'])
+
+        aliases = ','.join(sorted(product_aliases))
+        units = int(m["units"])
+        price = d['price']
+
+        # Apply a 30% margin and divide by the number of units per sold packaging.
+        unit_price = price * our_margin / units
+        # Round up to 5ct.
+        unit_price = (unit_price * 20).quantize(Decimal('1'), rounding=ROUND_UP) / 20
+
+        lines_out.append(f'{aliases}\t{unit_price:.2f}\t{d["name"]}  # {m["sku"]} {units}x')
+
+    return '\n'.join(lines_out)
diff --git a/inflatinator/scrapers.py b/inflatinator/scrapers.py
new file mode 100644
index 0000000..78c43ea
--- /dev/null
+++ b/inflatinator/scrapers.py
@@ -0,0 +1,37 @@
+from decimal import Decimal
+from pyquery import PyQuery as pq
+import json
+import re
+import subprocess
+
+
+def get(url):
+    compl = subprocess.run(['links', '-source', url], capture_output=True)
+    return compl.stdout
+
+
+def ah_get_by_sku(ah_sku):
+    assert re.match('^wi\d+$', ah_sku)
+
+    html_src = get(f'https://www.ah.nl/producten/product/{ah_sku}')
+    doc = pq(html_src)
+
+    ld_jsons = doc('script[type="application/ld+json"]')
+    for j in ld_jsons:
+        schema = json.loads(j.text)
+        if schema['@type'] == 'Product' and schema['sku'] == ah_sku:
+            break
+    else:
+        raise Exception(f'ah.nl returned no JSON metadata for SKU {ah_sku}')
+
+    name = schema['name']
+    ean = schema['gtin13']
+    sku = schema['sku']
+    price = Decimal(schema['offers']['price'])
+
+    return {
+        'name': name,
+        'price': price,
+        'ean': ean,
+        'sku': sku,
+    }